WRT54GS V2.0 QoS applicability?

Posts: 1715 · Posted: Thu Jul 06, 2006 3:33 pm Post subject:

Maestro wrote:

navydavy2001 wrote:

We're not talking about a fast changing corporate network or Internet Cafe. Home users don't have near as many changes to their network, and unless your service provider makes a mjor change, you CAN forget it. I ran my QOS settings for over a year with no changes, and had no problems. I had no ISP changes within that year, and added one new PC.

There are times that I have a lot of large packets moving in the network where LEMs are necessary to reduce jitter. Other times, I may be using TCP applications that are best served by using RED on them and PQ on the Voip. In other circumstances, I need to use other QoS techniques for optimum performance. This is on my home network.

When I travel, I need to evaluate the network to decide which QoS techniques that I am able to use, and which will help.

QoS is a large and complex subject, not a simple setting in a single network device.

I'm with NavyDavy on this one. Unless I'm working on something HIGHLY time-critical that would represent a major financial loss if it didn't go off perfectly (like a TV broadcast over 'net, yes, broadcast TV networks sometimes send their feeds via the internet), I just leave it alone at a general setting which keeps the Voip & audio streams happy at the expense of web surfing & FTP access.

Different strokes for different folks. Basically, I need to get work done & don't have time to screw around with my router & firewall every 5 minutes. I've configured my QoS to operate best with the normal traffic which I generate. I've accepted the fact that the changes I've made mean less-than-optimal performance in terms of speed, but work perfectly for regulating the flow of traffic out of my pipe. Could I get better speeds if I constantly tweaked? Sure. BUT more bandwith is cheaper than my time, so I buy more bandwidth as needed. It's more of a philosophical thing. I have a few networks w/matching servers out in the field which I've not touched in almost 3 years. I'll probably overhaul them later on this year, as I don't trust hardware for mission-critical apps after 3 years of 24/7 operation, but they've been running the five 9's ever since they were put into service. The less my clients see me in their office is a good thing -- it means I've done my job right.

Posts: 72 · Full Forum Member Joined: Jun 15, 2006 Posts: 72

I don't screw around with my router ever 5 minutes, but, for example, sometimes I need to use both Vonage lines, NetMeeting, and access things over the Internet at the same time. Unfortunately, most service providers have an outbound pipe that is too small for this to work properly without using a different QoS scheme than I normally use.

I know someone that dumped Vonage because she assumed that some QoS setting in a single device would fix everything, but she really needed to look at QoS which is an end-to-end process on your network. She had problems like losing a call when she printed. Changing an outbound PQ on a router won't do anything for that.

QoS needs to take the entire network (all devices, users, and traffic) into account to properly work.

Posts: 10 · Vonage Forum Associate Joined: Jul 02, 2006 Posts: 10

I had some interesting (better than expected) results in setting QoS for the first time today. In all the Voip speed tests I've done recently, I consistently got 375K up, both for wired and wireless connections. Throughout the day, I consistently got a reported 99% QOS (meaning consistent delivery of ~375K up). Based on that, I manually set the upload bandwidth to 330K, a little less than 90% of the sustained number. I decided to set WIRED QoS on a port basis because it was easy to do, setting port 3 on the WRT54GS (where my VTA is connected) to high, while leaving everything else with the default (LOW) setting. I did nothing with the Wireless QoS settings. After changing the wired QoS settings as described above, the test Voip calls (done to a voicemail box on a POTS phone) were very good quality. I performed Voip speed tests during the calls and was not surprised when they throttled back to about 250K upload throughput for the WIRED connection. The QoS was working as designed. The surprise: the Voip speed test when performed for the WIRELESS connection (while a Voip call was being made) ALSO throttled back to about 250K upload. This seems to indicate that the QoS algorithm used by the WRT54GS to parcel out the upload bandwidth for WIRED connections also takes into account the settings for the WIRELESS QoS. Anybody know different?

________________________________________________________________________________________
My setup: RoadRunner cable modem (4Mb down/375Kb up) <= WRT54GS 2.0/V4.71.1 <= DLink VTA <= Uniden TRU9485 5.8Ghz. Usually one laptop connected via wireless with Lotus Notes primary upload contender.